Facial recognition systems are vulnerable to physical (e.g., printed photos) and digital (e.g., DeepFake) face attacks. Existing methods struggle to simultaneously detect physical and digital attacks due to: 1) significant intra-class variations between these attack types, and 2) the inadequacy of spatial information alone to comprehensively capture live and fake cues. To address these issues, we propose a unified attack detection model termed Frequency-Aware and Attack-Agnostic CLIP (FA³-CLIP), which introduces attack-agnostic prompt learning to express generic live and fake cues derived from the fusion of spatial and frequency features, enabling unified detection of live faces and all categories of attacks. Specifically, the attack-agnostic prompt module generates generic live and fake prompts within the language branch to extract corresponding generic representations from both live and fake faces, guiding the model to learn a unified feature space for unified attack detection. Meanwhile, the module adaptively generates the live/fake conditional bias from the original spatial and frequency information to optimize the generic prompts accordingly, reducing the impact of intra-class variations. We further propose a dual-stream cues fusion framework in the vision branch, which leverages frequency information to complement subtle cues that are difficult to capture in the spatial domain. In addition, a frequency compression block is utilized in the frequency stream, which reduces redundancy in frequency features while preserving the diversity of crucial cues. We also establish new challenging protocols to facilitate unified face attack detection effectiveness. Experimental results on multiple benchmarks demonstrate that FA³-CLIP significantly improves performance, reducing ACER by over 1.2% on UniAttackData, and increasing AUC by more than 3% as well as reducing EER by over 4% on the JFSFDB dataset. The source code is available at https://github.com/YongzeLi/FA3-CLIP

Frequent user data breaches and misuse incidents highlight the flaws in current identity management systems. This study proposes a blockchain-based, peer-supervised self-sovereign identity (SSI) generation and privacy protection technology. Our approach creates unique digital identities on the blockchain, enabling secure cross-domain recognition and data sharing and satisfying the essential users’ requirements for SSI. Compared to existing SSI solutions, our approach has the practical advantages of less implementation cost, ease of users’ understanding and agreement, and better possibility of being soon adopted by current society and legal systems. The key innovative technical features include (1) using a zero-knowledge proof technology to ensure data remain “usable but invisible”, mitigating data breach risks; (2) introducing a peer review mechanism among service providers to prevent excessive data requests and misuse; and (3) implementing a comprehensive multi-party supervision system to audit all involved parties and prevent misconduct.

In multimodal sentiment analysis, a significant challenge lies in quantifying the contribution of each modality and achieving effective modality fusion. This paper presents a Hierarchical Gating-Driven Transformer Fusion Model (HGTFM), which effectively achieves multimodal data fusion through an advanced transformer architecture and a Hierarchical Gated Fusion Module (HGFM). The model first encodes audio and video features using LSTM and CNN, respectively, while employing BERT to encode text features. This approach effectively enhances the model ability to manage long-range temporal dependencies. Furthermore, the model introduces HGFM, which dynamically integrates multimodal features across different levels after processing by the modality interaction network (MIN), thus intelligently adjusting the contribution weights of each modality. In particular, the model incorporates a Unimodal Label Generation Module (ULGM), which considers the differences between the modalities. This enables HGFM to more accurately identify the contributions of samples with significant modal variations, thus optimizing the module potential. The experimental results on several established benchmark datasets demonstrate that HGTFM achieves comparable performance improvements in multimodal sentiment analysis tasks.

Temperature climate is an essential component of weather forecasting and is vital in predicting future weather patterns. Accurate temperature predictions can assist individuals and organizations in preparing for potential weather-related events such as heat waves or cold snaps. However, achieving precise temperature predictions necessitates thoroughly comprehending the underlying factors influencing climate patterns. The study utilized two models, LSTM and DLSTM, to forecast daily air temperature using 1097 data points gathered from central and southern regions of Tabriz city of Iran in Asia from 2017 to 2019. The results indicated that the proposed model had a high accuracy rate for predicting daily air temperature for test data, with RMSE_DLSTM = 0.08 °C and R-Square_DLSTM = 0.99. The DLSTM algorithm is known for its high speed, accuracy, time series prediction, noise reduction capabilities for data, the large volume of data processing, and improved performance of predicted data. In summary, while both LSTM and DLSTM are used for predicting data points, DLSTM is a more advanced version that includes multiple layers of memory cells and is better suited for handling complex sequences of events.

Multimodal sentiment analysis (MSA) faces key challenges such as incomplete modality inputs, long-range temporal dependencies, and suboptimal fusion strategies. To address these, we propose MGMR-Net, a Mamba-guided multimodal reconstruction and fusion network that integrates modality-aware reconstruction with text-centric fusion within an efficient state-space modeling framework. MGMR-Net consists of two core components: the Mamba-collaborative fusion module, which utilizes a two-stage selective state-space mechanism for fine-grained cross-modal alignment and hierarchical temporal integration, and the Mamba-enhanced reconstruction module, which employs continuous-time recurrence and dynamic gating to accurately recover corrupted or missing modality features. The entire network is jointly optimized via a unified multi-task loss, enabling simultaneous learning of discriminative features for sentiment prediction and reconstructive features for modality recovery. Extensive experiments on CMU-MOSI, CMU-MOSEI, and CH-SIMS datasets demonstrate that MGMR-Net consistently outperforms several baseline methods under both complete and missing modality settings, achieving superior accuracy, robustness, and generalization.